diff options
Diffstat (limited to 'drivers/net')
25 files changed, 392 insertions, 200 deletions
diff --git a/drivers/net/ethernet/amd/au1000_eth.c b/drivers/net/ethernet/amd/au1000_eth.c index a78e4c136959..31c48a7ac2b6 100644 --- a/drivers/net/ethernet/amd/au1000_eth.c +++ b/drivers/net/ethernet/amd/au1000_eth.c @@ -89,6 +89,124 @@ MODULE_DESCRIPTION(DRV_DESC); MODULE_LICENSE("GPL"); MODULE_VERSION(DRV_VERSION); +/* AU1000 MAC registers and bits */ +#define MAC_CONTROL 0x0 +# define MAC_RX_ENABLE (1 << 2) +# define MAC_TX_ENABLE (1 << 3) +# define MAC_DEF_CHECK (1 << 5) +# define MAC_SET_BL(X) (((X) & 0x3) << 6) +# define MAC_AUTO_PAD (1 << 8) +# define MAC_DISABLE_RETRY (1 << 10) +# define MAC_DISABLE_BCAST (1 << 11) +# define MAC_LATE_COL (1 << 12) +# define MAC_HASH_MODE (1 << 13) +# define MAC_HASH_ONLY (1 << 15) +# define MAC_PASS_ALL (1 << 16) +# define MAC_INVERSE_FILTER (1 << 17) +# define MAC_PROMISCUOUS (1 << 18) +# define MAC_PASS_ALL_MULTI (1 << 19) +# define MAC_FULL_DUPLEX (1 << 20) +# define MAC_NORMAL_MODE 0 +# define MAC_INT_LOOPBACK (1 << 21) +# define MAC_EXT_LOOPBACK (1 << 22) +# define MAC_DISABLE_RX_OWN (1 << 23) +# define MAC_BIG_ENDIAN (1 << 30) +# define MAC_RX_ALL (1 << 31) +#define MAC_ADDRESS_HIGH 0x4 +#define MAC_ADDRESS_LOW 0x8 +#define MAC_MCAST_HIGH 0xC +#define MAC_MCAST_LOW 0x10 +#define MAC_MII_CNTRL 0x14 +# define MAC_MII_BUSY (1 << 0) +# define MAC_MII_READ 0 +# define MAC_MII_WRITE (1 << 1) +# define MAC_SET_MII_SELECT_REG(X) (((X) & 0x1f) << 6) +# define MAC_SET_MII_SELECT_PHY(X) (((X) & 0x1f) << 11) +#define MAC_MII_DATA 0x18 +#define MAC_FLOW_CNTRL 0x1C +# define MAC_FLOW_CNTRL_BUSY (1 << 0) +# define MAC_FLOW_CNTRL_ENABLE (1 << 1) +# define MAC_PASS_CONTROL (1 << 2) +# define MAC_SET_PAUSE(X) (((X) & 0xffff) << 16) +#define MAC_VLAN1_TAG 0x20 +#define MAC_VLAN2_TAG 0x24 + +/* Ethernet Controller Enable */ +# define MAC_EN_CLOCK_ENABLE (1 << 0) +# define MAC_EN_RESET0 (1 << 1) +# define MAC_EN_TOSS (0 << 2) +# define MAC_EN_CACHEABLE (1 << 3) +# define MAC_EN_RESET1 (1 << 4) +# define MAC_EN_RESET2 (1 << 5) +# define MAC_DMA_RESET (1 << 6) + +/* Ethernet Controller DMA Channels */ +/* offsets from MAC_TX_RING_ADDR address */ +#define MAC_TX_BUFF0_STATUS 0x0 +# define TX_FRAME_ABORTED (1 << 0) +# define TX_JAB_TIMEOUT (1 << 1) +# define TX_NO_CARRIER (1 << 2) +# define TX_LOSS_CARRIER (1 << 3) +# define TX_EXC_DEF (1 << 4) +# define TX_LATE_COLL_ABORT (1 << 5) +# define TX_EXC_COLL (1 << 6) +# define TX_UNDERRUN (1 << 7) +# define TX_DEFERRED (1 << 8) +# define TX_LATE_COLL (1 << 9) +# define TX_COLL_CNT_MASK (0xF << 10) +# define TX_PKT_RETRY (1 << 31) +#define MAC_TX_BUFF0_ADDR 0x4 +# define TX_DMA_ENABLE (1 << 0) +# define TX_T_DONE (1 << 1) +# define TX_GET_DMA_BUFFER(X) (((X) >> 2) & 0x3) +#define MAC_TX_BUFF0_LEN 0x8 +#define MAC_TX_BUFF1_STATUS 0x10 +#define MAC_TX_BUFF1_ADDR 0x14 +#define MAC_TX_BUFF1_LEN 0x18 +#define MAC_TX_BUFF2_STATUS 0x20 +#define MAC_TX_BUFF2_ADDR 0x24 +#define MAC_TX_BUFF2_LEN 0x28 +#define MAC_TX_BUFF3_STATUS 0x30 +#define MAC_TX_BUFF3_ADDR 0x34 +#define MAC_TX_BUFF3_LEN 0x38 + +/* offsets from MAC_RX_RING_ADDR */ +#define MAC_RX_BUFF0_STATUS 0x0 +# define RX_FRAME_LEN_MASK 0x3fff +# define RX_WDOG_TIMER (1 << 14) +# define RX_RUNT (1 << 15) +# define RX_OVERLEN (1 << 16) +# define RX_COLL (1 << 17) +# define RX_ETHER (1 << 18) +# define RX_MII_ERROR (1 << 19) +# define RX_DRIBBLING (1 << 20) +# define RX_CRC_ERROR (1 << 21) +# define RX_VLAN1 (1 << 22) +# define RX_VLAN2 (1 << 23) +# define RX_LEN_ERROR (1 << 24) +# define RX_CNTRL_FRAME (1 << 25) +# define RX_U_CNTRL_FRAME (1 << 26) +# define RX_MCAST_FRAME (1 << 27) +# define RX_BCAST_FRAME (1 << 28) +# define RX_FILTER_FAIL (1 << 29) +# define RX_PACKET_FILTER (1 << 30) +# define RX_MISSED_FRAME (1 << 31) + +# define RX_ERROR (RX_WDOG_TIMER | RX_RUNT | RX_OVERLEN | \ + RX_COLL | RX_MII_ERROR | RX_CRC_ERROR | \ + RX_LEN_ERROR | RX_U_CNTRL_FRAME | RX_MISSED_FRAME) +#define MAC_RX_BUFF0_ADDR 0x4 +# define RX_DMA_ENABLE (1 << 0) +# define RX_T_DONE (1 << 1) +# define RX_GET_DMA_BUFFER(X) (((X) >> 2) & 0x3) +# define RX_SET_BUFF_ADDR(X) ((X) & 0xffffffc0) +#define MAC_RX_BUFF1_STATUS 0x10 +#define MAC_RX_BUFF1_ADDR 0x14 +#define MAC_RX_BUFF2_STATUS 0x20 +#define MAC_RX_BUFF2_ADDR 0x24 +#define MAC_RX_BUFF3_STATUS 0x30 +#define MAC_RX_BUFF3_ADDR 0x34 + /* * Theory of operation * @@ -152,10 +270,12 @@ static void au1000_enable_mac(struct net_device *dev, int force_reset) if (force_reset || (!aup->mac_enabled)) { writel(MAC_EN_CLOCK_ENABLE, aup->enable); - au_sync_delay(2); + wmb(); /* drain writebuffer */ + mdelay(2); writel((MAC_EN_RESET0 | MAC_EN_RESET1 | MAC_EN_RESET2 | MAC_EN_CLOCK_ENABLE), aup->enable); - au_sync_delay(2); + wmb(); /* drain writebuffer */ + mdelay(2); aup->mac_enabled = 1; } @@ -273,7 +393,8 @@ static void au1000_hard_stop(struct net_device *dev) reg = readl(&aup->mac->control); reg &= ~(MAC_RX_ENABLE | MAC_TX_ENABLE); writel(reg, &aup->mac->control); - au_sync_delay(10); + wmb(); /* drain writebuffer */ + mdelay(10); } static void au1000_enable_rx_tx(struct net_device *dev) @@ -286,7 +407,8 @@ static void au1000_enable_rx_tx(struct net_device *dev) reg = readl(&aup->mac->control); reg |= (MAC_RX_ENABLE | MAC_TX_ENABLE); writel(reg, &aup->mac->control); - au_sync_delay(10); + wmb(); /* drain writebuffer */ + mdelay(10); } static void @@ -336,7 +458,8 @@ au1000_adjust_link(struct net_device *dev) reg |= MAC_DISABLE_RX_OWN; } writel(reg, &aup->mac->control); - au_sync_delay(1); + wmb(); /* drain writebuffer */ + mdelay(1); au1000_enable_rx_tx(dev); aup->old_duplex = phydev->duplex; @@ -500,9 +623,11 @@ static void au1000_reset_mac_unlocked(struct net_device *dev) au1000_hard_stop(dev); writel(MAC_EN_CLOCK_ENABLE, aup->enable); - au_sync_delay(2); + wmb(); /* drain writebuffer */ + mdelay(2); writel(0, aup->enable); - au_sync_delay(2); + wmb(); /* drain writebuffer */ + mdelay(2); aup->tx_full = 0; for (i = 0; i < NUM_RX_DMA; i++) { @@ -652,7 +777,7 @@ static int au1000_init(struct net_device *dev) for (i = 0; i < NUM_RX_DMA; i++) aup->rx_dma_ring[i]->buff_stat |= RX_DMA_ENABLE; - au_sync(); + wmb(); /* drain writebuffer */ control = MAC_RX_ENABLE | MAC_TX_ENABLE; #ifndef CONFIG_CPU_LITTLE_ENDIAN @@ -669,7 +794,7 @@ static int au1000_init(struct net_device *dev) writel(control, &aup->mac->control); writel(0x8100, &aup->mac->vlan1_tag); /* activate vlan support */ - au_sync(); + wmb(); /* drain writebuffer */ spin_unlock_irqrestore(&aup->lock, flags); return 0; @@ -760,7 +885,7 @@ static int au1000_rx(struct net_device *dev) } prxd->buff_stat = (u32)(pDB->dma_addr | RX_DMA_ENABLE); aup->rx_head = (aup->rx_head + 1) & (NUM_RX_DMA - 1); - au_sync(); + wmb(); /* drain writebuffer */ /* next descriptor */ prxd = aup->rx_dma_ring[aup->rx_head]; @@ -808,7 +933,7 @@ static void au1000_tx_ack(struct net_device *dev) au1000_update_tx_stats(dev, ptxd->status); ptxd->buff_stat &= ~TX_T_DONE; ptxd->len = 0; - au_sync(); + wmb(); /* drain writebuffer */ aup->tx_tail = (aup->tx_tail + 1) & (NUM_TX_DMA - 1); ptxd = aup->tx_dma_ring[aup->tx_tail]; @@ -939,7 +1064,7 @@ static netdev_tx_t au1000_tx(struct sk_buff *skb, struct net_device *dev) ps->tx_bytes += ptxd->len; ptxd->buff_stat = pDB->dma_addr | TX_DMA_ENABLE; - au_sync(); + wmb(); /* drain writebuffer */ dev_kfree_skb(skb); aup->tx_head = (aup->tx_head + 1) & (NUM_TX_DMA - 1); return NETDEV_TX_OK; diff --git a/drivers/net/ethernet/amd/pcnet32.c b/drivers/net/ethernet/amd/pcnet32.c index e7cc9174e364..4a8fdc4721d5 100644 --- a/drivers/net/ethernet/amd/pcnet32.c +++ b/drivers/net/ethernet/amd/pcnet32.c @@ -481,37 +481,32 @@ static void pcnet32_realloc_tx_ring(struct net_device *dev, dma_addr_t *new_dma_addr_list; struct pcnet32_tx_head *new_tx_ring; struct sk_buff **new_skb_list; + unsigned int entries = BIT(size); pcnet32_purge_tx_ring(dev); - new_tx_ring = pci_alloc_consistent(lp->pci_dev, - sizeof(struct pcnet32_tx_head) * - (1 << size), - &new_ring_dma_addr); - if (new_tx_ring == NULL) { - netif_err(lp, drv, dev, "Consistent memory allocation failed\n"); + new_tx_ring = + pci_zalloc_consistent(lp->pci_dev, + sizeof(struct pcnet32_tx_head) * entries, + &new_ring_dma_addr); + if (new_tx_ring == NULL) return; - } - memset(new_tx_ring, 0, sizeof(struct pcnet32_tx_head) * (1 << size)); - new_dma_addr_list = kcalloc(1 << size, sizeof(dma_addr_t), - GFP_ATOMIC); + new_dma_addr_list = kcalloc(entries, sizeof(dma_addr_t), GFP_ATOMIC); if (!new_dma_addr_list) goto free_new_tx_ring; - new_skb_list = kcalloc(1 << size, sizeof(struct sk_buff *), - GFP_ATOMIC); + new_skb_list = kcalloc(entries, sizeof(struct sk_buff *), GFP_ATOMIC); if (!new_skb_list) goto free_new_lists; kfree(lp->tx_skbuff); kfree(lp->tx_dma_addr); pci_free_consistent(lp->pci_dev, - sizeof(struct pcnet32_tx_head) * - lp->tx_ring_size, lp->tx_ring, - lp->tx_ring_dma_addr); + sizeof(struct pcnet32_tx_head) * lp->tx_ring_size, + lp->tx_ring, lp->tx_ring_dma_addr); - lp->tx_ring_size = (1 << size); + lp->tx_ring_size = entries; lp->tx_mod_mask = lp->tx_ring_size - 1; lp->tx_len_bits = (size << 12); lp->tx_ring = new_tx_ring; @@ -524,8 +519,7 @@ free_new_lists: kfree(new_dma_addr_list); free_new_tx_ring: pci_free_consistent(lp->pci_dev, - sizeof(struct pcnet32_tx_head) * - (1 << size), + sizeof(struct pcnet32_tx_head) * entries, new_tx_ring, new_ring_dma_addr); } @@ -549,17 +543,14 @@ static void pcnet32_realloc_rx_ring(struct net_device *dev, struct pcnet32_rx_head *new_rx_ring; struct sk_buff **new_skb_list; int new, overlap; - unsigned int entries = 1 << size; + unsigned int entries = BIT(size); - new_rx_ring = pci_alloc_consistent(lp->pci_dev, - sizeof(struct pcnet32_rx_head) * - entries, - &new_ring_dma_addr); - if (new_rx_ring == NULL) { - netif_err(lp, drv, dev, "Consistent memory allocation failed\n"); + new_rx_ring = + pci_zalloc_consistent(lp->pci_dev, + sizeof(struct pcnet32_rx_head) * entries, + &new_ring_dma_addr); + if (new_rx_ring == NULL) return; - } - memset(new_rx_ring, 0, sizeof(struct pcnet32_rx_head) * entries); new_dma_addr_list = kcalloc(entries, sizeof(dma_addr_t), GFP_ATOMIC); if (!new_dma_addr_list) diff --git a/drivers/net/ethernet/atheros/atl1e/atl1e_main.c b/drivers/net/ethernet/atheros/atl1e/atl1e_main.c index 4345332533ad..316e0c3fe048 100644 --- a/drivers/net/ethernet/atheros/atl1e/atl1e_main.c +++ b/drivers/net/ethernet/atheros/atl1e/atl1e_main.c @@ -831,17 +831,14 @@ static int atl1e_setup_ring_resources(struct atl1e_adapter *adapter) /* real ring DMA buffer */ size = adapter->ring_size; - adapter->ring_vir_addr = pci_alloc_consistent(pdev, - adapter->ring_size, &adapter->ring_dma); - + adapter->ring_vir_addr = pci_zalloc_consistent(pdev, adapter->ring_size, + &adapter->ring_dma); if (adapter->ring_vir_addr == NULL) { netdev_err(adapter->netdev, "pci_alloc_consistent failed, size = D%d\n", size); return -ENOMEM; } - memset(adapter->ring_vir_addr, 0, adapter->ring_size); - rx_page_desc = rx_ring->rx_page_desc; /* Init TPD Ring */ diff --git a/drivers/net/ethernet/broadcom/tg3.c b/drivers/net/ethernet/broadcom/tg3.c index a3dd5dc64f4c..4296b3d26f02 100644 --- a/drivers/net/ethernet/broadcom/tg3.c +++ b/drivers/net/ethernet/broadcom/tg3.c @@ -14093,8 +14093,9 @@ static struct rtnl_link_stats64 *tg3_get_stats64(struct net_device *dev, spin_lock_bh(&tp->lock); if (!tp->hw_stats) { + *stats = tp->net_stats_prev; spin_unlock_bh(&tp->lock); - return &tp->net_stats_prev; + return stats; } tg3_get_nstats(tp, stats); diff --git a/drivers/net/ethernet/cisco/enic/vnic_dev.c b/drivers/net/ethernet/cisco/enic/vnic_dev.c index 5abc496bcf29..37472ce4fac3 100644 --- a/drivers/net/ethernet/cisco/enic/vnic_dev.c +++ b/drivers/net/ethernet/cisco/enic/vnic_dev.c @@ -432,14 +432,12 @@ int vnic_dev_fw_info(struct vnic_dev *vdev, int err = 0; if (!vdev->fw_info) { - vdev->fw_info = pci_alloc_consistent(vdev->pdev, - sizeof(struct vnic_devcmd_fw_info), - &vdev->fw_info_pa); + vdev->fw_info = pci_zalloc_consistent(vdev->pdev, + sizeof(struct vnic_devcmd_fw_info), + &vdev->fw_info_pa); if (!vdev->fw_info) return -ENOMEM; - memset(vdev->fw_info, 0, sizeof(struct vnic_devcmd_fw_info)); - a0 = vdev->fw_info_pa; a1 = sizeof(struct vnic_devcmd_fw_info); diff --git a/drivers/net/ethernet/freescale/fs_enet/mac-fec.c b/drivers/net/ethernet/freescale/fs_enet/mac-fec.c index fc5413488496..1eedfba2ad3c 100644 --- a/drivers/net/ethernet/freescale/fs_enet/mac-fec.c +++ b/drivers/net/ethernet/freescale/fs_enet/mac-fec.c @@ -41,7 +41,6 @@ #ifdef CONFIG_8xx #include <asm/8xx_immap.h> #include <asm/pgtable.h> -#include <asm/mpc8xx.h> #include <asm/cpm1.h> #endif diff --git a/drivers/net/ethernet/freescale/fs_enet/mac-scc.c b/drivers/net/ethernet/freescale/fs_enet/mac-scc.c index b4bf02f57d43..90b3b19b7cd3 100644 --- a/drivers/net/ethernet/freescale/fs_enet/mac-scc.c +++ b/drivers/net/ethernet/freescale/fs_enet/mac-scc.c @@ -40,7 +40,6 @@ #ifdef CONFIG_8xx #include <asm/8xx_immap.h> #include <asm/pgtable.h> -#include <asm/mpc8xx.h> #include <asm/cpm1.h> #endif diff --git a/drivers/net/ethernet/marvell/sky2.c b/drivers/net/ethernet/marvell/sky2.c index 69693384b58c..59915144aabb 100644 --- a/drivers/net/ethernet/marvell/sky2.c +++ b/drivers/net/ethernet/marvell/sky2.c @@ -1622,11 +1622,10 @@ static int sky2_alloc_buffers(struct sky2_port *sky2) if (!sky2->tx_ring) goto nomem; - sky2->rx_le = pci_alloc_consistent(hw->pdev, RX_LE_BYTES, - &sky2->rx_le_map); + sky2->rx_le = pci_zalloc_consistent(hw->pdev, RX_LE_BYTES, + &sky2->rx_le_map); if (!sky2->rx_le) goto nomem; - memset(sky2->rx_le, 0, RX_LE_BYTES); sky2->rx_ring = kcalloc(sky2->rx_pending, sizeof(struct rx_ring_info), GFP_KERNEL); diff --git a/drivers/net/ethernet/micrel/ksz884x.c b/drivers/net/ethernet/micrel/ksz884x.c index 064a48d0c368..cd5f106306d9 100644 --- a/drivers/net/ethernet/micrel/ksz884x.c +++ b/drivers/net/ethernet/micrel/ksz884x.c @@ -4409,14 +4409,13 @@ static int ksz_alloc_desc(struct dev_info *adapter) DESC_ALIGNMENT; adapter->desc_pool.alloc_virt = - pci_alloc_consistent( - adapter->pdev, adapter->desc_pool.alloc_size, - &adapter->desc_pool.dma_addr); + pci_zalloc_consistent(adapter->pdev, + adapter->desc_pool.alloc_size, + &adapter->desc_pool.dma_addr); if (adapter->desc_pool.alloc_virt == NULL) { adapter->desc_pool.alloc_size = 0; return 1; } - memset(adapter->desc_pool.alloc_virt, 0, adapter->desc_pool.alloc_size); /* Align to the next cache line boundary. */ offset = (((ulong) adapter->desc_pool.alloc_virt % DESC_ALIGNMENT) ? diff --git a/drivers/net/ethernet/myricom/myri10ge/myri10ge.c b/drivers/net/ethernet/myricom/myri10ge/myri10ge.c index 69c26f04d8ce..679db026f4be 100644 --- a/drivers/net/ethernet/myricom/myri10ge/myri10ge.c +++ b/drivers/net/ethernet/myricom/myri10ge/myri10ge.c @@ -873,6 +873,10 @@ static int myri10ge_dma_test(struct myri10ge_priv *mgp, int test_type) return -ENOMEM; dmatest_bus = pci_map_page(mgp->pdev, dmatest_page, 0, PAGE_SIZE, DMA_BIDIRECTIONAL); + if (unlikely(pci_dma_mapping_error(mgp->pdev, dmatest_bus))) { + __free_page(dmatest_page); + return -ENOMEM; + } /* Run a small DMA test. * The magic multipliers to the length tell the firmware @@ -1294,6 +1298,7 @@ myri10ge_alloc_rx_pages(struct myri10ge_priv *mgp, struct myri10ge_rx_buf *rx, int bytes, int watchdog) { struct page *page; + dma_addr_t bus; int idx; #if MYRI10GE_ALLOC_SIZE > 4096 int end_offset; @@ -1318,11 +1323,21 @@ myri10ge_alloc_rx_pages(struct myri10ge_priv *mgp, struct myri10ge_rx_buf *rx, rx->watchdog_needed = 1; return; } + + bus = pci_map_page(mgp->pdev, page, 0, + MYRI10GE_ALLOC_SIZE, + PCI_DMA_FROMDEVICE); + if (unlikely(pci_dma_mapping_error(mgp->pdev, bus))) { + __free_pages(page, MYRI10GE_ALLOC_ORDER); + if (rx->fill_cnt - rx->cnt < 16) + rx->watchdog_needed = 1; + return; + } + rx->page = page; rx->page_offset = 0; - rx->bus = pci_map_page(mgp->pdev, page, 0, - MYRI10GE_ALLOC_SIZE, - PCI_DMA_FROMDEVICE); + rx->bus = bus; + } rx->info[idx].page = rx->page; rx->info[idx].page_offset = rx->page_offset; @@ -2764,6 +2779,35 @@ myri10ge_submit_req(struct myri10ge_tx_buf *tx, struct mcp_kreq_ether_send *src, mb(); } +static void myri10ge_unmap_tx_dma(struct myri10ge_priv *mgp, + struct myri10ge_tx_buf *tx, int idx) +{ + unsigned int len; + int last_idx; + + /* Free any DMA resources we've alloced and clear out the skb slot */ + last_idx = (idx + 1) & tx->mask; + idx = tx->req & tx->mask; + do { + len = dma_unmap_len(&tx->info[idx], len); + if (len) { + if (tx->info[idx].skb != NULL) + pci_unmap_single(mgp->pdev, + dma_unmap_addr(&tx->info[idx], + bus), len, + PCI_DMA_TODEVICE); + else + pci_unmap_page(mgp->pdev, + dma_unmap_addr(&tx->info[idx], + bus), len, + PCI_DMA_TODEVICE); + dma_unmap_len_set(&tx->info[idx], len, 0); + tx->info[idx].skb = NULL; + } + idx = (idx + 1) & tx->mask; + } while (idx != last_idx); +} + /* * Transmit a packet. We need to split the packet so that a single * segment does not cross myri10ge->tx_boundary, so this makes segment @@ -2787,7 +2831,7 @@ static netdev_tx_t myri10ge_xmit(struct sk_buff *skb, u32 low; __be32 high_swapped; unsigned int len; - int idx, last_idx, avail, frag_cnt, frag_idx, count, mss, max_segments; + int idx, avail, frag_cnt, frag_idx, count, mss, max_segments; u16 pseudo_hdr_offset, cksum_offset, queue; int cum_len, seglen, boundary, rdma_count; u8 flags, odd_flag; @@ -2884,9 +2928,12 @@ again: /* map the skb for DMA */ len = skb_headlen(skb); + bus = pci_map_single(mgp->pdev, skb->data, len, PCI_DMA_TODEVICE); + if (unlikely(pci_dma_mapping_error(mgp->pdev, bus))) + goto drop; + idx = tx->req & tx->mask; tx->info[idx].skb = skb; - bus = pci_map_single(mgp->pdev, skb->data, len, PCI_DMA_TODEVICE); dma_unmap_addr_set(&tx->info[idx], bus, bus); dma_unmap_len_set(&tx->info[idx], len, len); @@ -2985,12 +3032,16 @@ again: break; /* map next fragment for DMA */ - idx = (count + tx->req) & tx->mask; frag = &skb_shinfo(skb)->frags[frag_idx]; frag_idx++; len = skb_frag_size(frag); bus = skb_frag_dma_map(&mgp->pdev->dev, frag, 0, len, DMA_TO_DEVICE); + if (unlikely(pci_dma_mapping_error(mgp->pdev, bus))) { + myri10ge_unmap_tx_dma(mgp, tx, idx); + goto drop; + } + idx = (count + tx->req) & tx->mask; dma_unmap_addr_set(&tx->info[idx], bus, bus); dma_unmap_len_set(&tx->info[idx], len, len); } @@ -3021,31 +3072,8 @@ again: return NETDEV_TX_OK; abort_linearize: - /* Free any DMA resources we've alloced and clear out the skb - * slot so as to not trip up assertions, and to avoid a - * double-free if linearizing fails */ + myri10ge_unmap_tx_dma(mgp, tx, idx); - last_idx = (idx + 1) & tx->mask; - idx = tx->req & tx->mask; - tx->info[idx].skb = NULL; - do { - len = dma_unmap_len(&tx->info[idx], len); - if (len) { - if (tx->info[idx].skb != NULL) - pci_unmap_single(mgp->pdev, - dma_unmap_addr(&tx->info[idx], - bus), len, - PCI_DMA_TODEVICE); - else - pci_unmap_page(mgp->pdev, - dma_unmap_addr(&tx->info[idx], - bus), len, - PCI_DMA_TODEVICE); - dma_unmap_len_set(&tx->info[idx], len, 0); - tx->info[idx].skb = NULL; - } - idx = (idx + 1) & tx->mask; - } while (idx != last_idx); if (skb_is_gso(skb)) { netdev_err(mgp->dev, "TSO but wanted to linearize?!?!?\n"); goto drop; diff --git a/drivers/net/ethernet/qlogic/netxen/netxen_nic_ctx.c b/drivers/net/ethernet/qlogic/netxen/netxen_nic_ctx.c index 6f6be57f4690..b8d5270359cd 100644 --- a/drivers/net/ethernet/qlogic/netxen/netxen_nic_ctx.c +++ b/drivers/net/ethernet/qlogic/netxen/netxen_nic_ctx.c @@ -129,14 +129,12 @@ netxen_get_minidump_template(struct netxen_adapter *adapter) return NX_RCODE_INVALID_ARGS; } - addr = pci_alloc_consistent(adapter->pdev, size, &md_template_addr); - + addr = pci_zalloc_consistent(adapter->pdev, size, &md_template_addr); if (!addr) { dev_err(&adapter->pdev->dev, "Unable to allocate dmable memory for template.\n"); return -ENOMEM; } - memset(addr, 0, size); memset(&cmd, 0, sizeof(cmd)); memset(&cmd.rsp, 1, sizeof(struct _cdrp_cmd)); cmd.req.cmd = NX_CDRP_CMD_GET_TEMP_HDR; diff --git a/drivers/net/ethernet/qlogic/qlge/qlge_main.c b/drivers/net/ethernet/qlogic/qlge/qlge_main.c index b40050e03a56..d836ace52277 100644 --- a/drivers/net/ethernet/qlogic/qlge/qlge_main.c +++ b/drivers/net/ethernet/qlogic/qlge/qlge_main.c @@ -2727,23 +2727,22 @@ static void ql_free_shadow_space(struct ql_adapter *qdev) static int ql_alloc_shadow_space(struct ql_adapter *qdev) { qdev->rx_ring_shadow_reg_area = - pci_alloc_consistent(qdev->pdev, - PAGE_SIZE, &qdev->rx_ring_shadow_reg_dma); + pci_zalloc_consistent(qdev->pdev, PAGE_SIZE, + &qdev->rx_ring_shadow_reg_dma); if (qdev->rx_ring_shadow_reg_area == NULL) { netif_err(qdev, ifup, qdev->ndev, "Allocation of RX shadow space failed.\n"); return -ENOMEM; } - memset(qdev->rx_ring_shadow_reg_area, 0, PAGE_SIZE); + qdev->tx_ring_shadow_reg_area = - pci_alloc_consistent(qdev->pdev, PAGE_SIZE, - &qdev->tx_ring_shadow_reg_dma); + pci_zalloc_consistent(qdev->pdev, PAGE_SIZE, + &qdev->tx_ring_shadow_reg_dma); if (qdev->tx_ring_shadow_reg_area == NULL) { netif_err(qdev, ifup, qdev->ndev, "Allocation of TX shadow space failed.\n"); goto err_wqp_sh_area; } - memset(qdev->tx_ring_shadow_reg_area, 0, PAGE_SIZE); return 0; err_wqp_sh_area: diff --git a/drivers/net/ethernet/sun/sunvnet.c b/drivers/net/ethernet/sun/sunvnet.c index d813bfb1a847..23c89ab5a6ad 100644 --- a/drivers/net/ethernet/sun/sunvnet.c +++ b/drivers/net/ethernet/sun/sunvnet.c @@ -32,6 +32,11 @@ MODULE_DESCRIPTION("Sun LDOM virtual network driver"); MODULE_LICENSE("GPL"); MODULE_VERSION(DRV_MODULE_VERSION); +/* Heuristic for the number of times to exponentially backoff and + * retry sending an LDC trigger when EAGAIN is encountered + */ +#define VNET_MAX_RETRIES 10 + /* Ordered from largest major to lowest */ static struct vio_version vnet_versions[] = { { .major = 1, .minor = 0 }, @@ -260,6 +265,7 @@ static int vnet_send_ack(struct vnet_port *port, struct vio_dring_state *dr, .state = vio_dring_state, }; int err, delay; + int retries = 0; hdr.seq = dr->snd_nxt; delay = 1; @@ -272,6 +278,13 @@ static int vnet_send_ack(struct vnet_port *port, struct vio_dring_state *dr, udelay(delay); if ((delay <<= 1) > 128) delay = 128; + if (retries++ > VNET_MAX_RETRIES) { + pr_info("ECONNRESET %x:%x:%x:%x:%x:%x\n", + port->raddr[0], port->raddr[1], + port->raddr[2], port->raddr[3], + port->raddr[4], port->raddr[5]); + err = -ECONNRESET; + } } while (err == -EAGAIN); return err; @@ -475,8 +488,9 @@ static int handle_mcast(struct vnet_port *port, void *msgbuf) return 0; } -static void maybe_tx_wakeup(struct vnet *vp) +static void maybe_tx_wakeup(unsigned long param) { + struct vnet *vp = (struct vnet *)param; struct net_device *dev = vp->dev; netif_tx_lock(dev); @@ -573,8 +587,13 @@ static void vnet_event(void *arg, int event) break; } spin_unlock(&vio->lock); + /* Kick off a tasklet to wake the queue. We cannot call + * maybe_tx_wakeup directly here because we could deadlock on + * netif_tx_lock() with dev_watchdog() + */ if (unlikely(tx_wakeup && err != -ECONNRESET)) - maybe_tx_wakeup(port->vp); + tasklet_schedule(&port->vp->vnet_tx_wakeup); + local_irq_restore(flags); } @@ -593,6 +612,7 @@ static int __vnet_tx_trigger(struct vnet_port *port) .end_idx = (u32) -1, }; int err, delay; + int retries = 0; hdr.seq = dr->snd_nxt; delay = 1; @@ -605,6 +625,8 @@ static int __vnet_tx_trigger(struct vnet_port *port) udelay(delay); if ((delay <<= 1) > 128) delay = 128; + if (retries++ > VNET_MAX_RETRIES) + break; } while (err == -EAGAIN); return err; @@ -691,7 +713,15 @@ static int vnet_start_xmit(struct sk_buff *skb, struct net_device *dev) memset(tx_buf+VNET_PACKET_SKIP+skb->len, 0, len - skb->len); } - d->hdr.ack = VIO_ACK_ENABLE; + /* We don't rely on the ACKs to free the skb in vnet_start_xmit(), + * thus it is safe to not set VIO_ACK_ENABLE for each transmission: + * the protocol itself does not require it as long as the peer + * sends a VIO_SUBTYPE_ACK for VIO_DRING_STOPPED. + * + * An ACK for every packet in the ring is expensive as the + * sending of LDC messages is slow and affects performance. + */ + d->hdr.ack = VIO_ACK_DISABLE; d->size = len; d->ncookies = port->tx_bufs[dr->prod].ncookies; for (i = 0; i < d->ncookies; i++) @@ -1046,6 +1076,7 @@ static struct vnet *vnet_new(const u64 *local_mac) vp = netdev_priv(dev); spin_lock_init(&vp->lock); + tasklet_init(&vp->vnet_tx_wakeup, maybe_tx_wakeup, (unsigned long)vp); vp->dev = dev; INIT_LIST_HEAD(&vp->port_list); @@ -1105,6 +1136,7 @@ static void vnet_cleanup(void) vp = list_first_entry(&vnet_list, struct vnet, list); list_del(&vp->list); dev = vp->dev; + tasklet_kill(&vp->vnet_tx_wakeup); /* vio_unregister_driver() should have cleaned up port_list */ BUG_ON(!list_empty(&vp->port_list)); unregister_netdev(dev); diff --git a/drivers/net/ethernet/sun/sunvnet.h b/drivers/net/ethernet/sun/sunvnet.h index d347a5bf24b0..de5c2c64996f 100644 --- a/drivers/net/ethernet/sun/sunvnet.h +++ b/drivers/net/ethernet/sun/sunvnet.h @@ -1,6 +1,8 @@ #ifndef _SUNVNET_H #define _SUNVNET_H +#include <linux/interrupt.h> + #define DESC_NCOOKIES(entry_size) \ ((entry_size) - sizeof(struct vio_net_desc)) @@ -78,6 +80,8 @@ struct vnet { struct list_head list; u64 local_mac; + + struct tasklet_struct vnet_tx_wakeup; }; #endif /* _SUNVNET_H */ diff --git a/drivers/net/irda/au1k_ir.c b/drivers/net/irda/au1k_ir.c index 5f91e3e01c04..aab2cf72d025 100644 --- a/drivers/net/irda/au1k_ir.c +++ b/drivers/net/irda/au1k_ir.c @@ -18,6 +18,7 @@ * with this program; if not, see <http://www.gnu.org/licenses/>. */ +#include <linux/clk.h> #include <linux/module.h> #include <linux/netdevice.h> #include <linux/interrupt.h> @@ -175,6 +176,7 @@ struct au1k_private { struct resource *ioarea; struct au1k_irda_platform_data *platdata; + struct clk *irda_clk; }; static int qos_mtt_bits = 0x07; /* 1 ms or more */ @@ -514,9 +516,39 @@ static irqreturn_t au1k_irda_interrupt(int dummy, void *dev_id) static int au1k_init(struct net_device *dev) { struct au1k_private *aup = netdev_priv(dev); - u32 enable, ring_address; + u32 enable, ring_address, phyck; + struct clk *c; int i; + c = clk_get(NULL, "irda_clk"); + if (IS_ERR(c)) + return PTR_ERR(c); + i = clk_prepare_enable(c); + if (i) { + clk_put(c); + return i; + } + + switch (clk_get_rate(c)) { + case 40000000: + phyck = IR_PHYCLK_40MHZ; + break; + case 48000000: + phyck = IR_PHYCLK_48MHZ; + break; + case 56000000: + phyck = IR_PHYCLK_56MHZ; + break; + case 64000000: + phyck = IR_PHYCLK_64MHZ; + break; + default: + clk_disable_unprepare(c); + clk_put(c); + return -EINVAL; + } + aup->irda_clk = c; + enable = IR_HC | IR_CE | IR_C; #ifndef CONFIG_CPU_LITTLE_ENDIAN enable |= IR_BE; @@ -545,7 +577,7 @@ static int au1k_init(struct net_device *dev) irda_write(aup, IR_RING_SIZE, (RING_SIZE_64 << 8) | (RING_SIZE_64 << 12)); - irda_write(aup, IR_CONFIG_2, IR_PHYCLK_48MHZ | IR_ONE_PIN); + irda_write(aup, IR_CONFIG_2, phyck | IR_ONE_PIN); irda_write(aup, IR_RING_ADDR_CMPR, 0); au1k_irda_set_speed(dev, 9600); @@ -619,6 +651,9 @@ static int au1k_irda_stop(struct net_device *dev) free_irq(aup->irq_tx, dev); free_irq(aup->irq_rx, dev); + clk_disable_unprepare(aup->irda_clk); + clk_put(aup->irda_clk); + return 0; } @@ -853,6 +888,7 @@ static int au1k_irda_probe(struct platform_device *pdev) struct au1k_private *aup; struct net_device *dev; struct resource *r; + struct clk *c; int err; dev = alloc_irdadev(sizeof(struct au1k_private)); @@ -886,6 +922,14 @@ static int au1k_irda_probe(struct platform_device *pdev) if (!aup->ioarea) goto out; + /* bail out early if clock doesn't exist */ + c = clk_get(NULL, "irda_clk"); + if (IS_ERR(c)) { + err = PTR_ERR(c); + goto out; + } + clk_put(c); + aup->iobase = ioremap_nocache(r->start, resource_size(r)); if (!aup->iobase) goto out2; diff --git a/drivers/net/irda/donauboe.c b/drivers/net/irda/donauboe.c index 768dfe9a9315..6d3e2093bf7f 100644 --- a/drivers/net/irda/donauboe.c +++ b/drivers/net/irda/donauboe.c @@ -1755,17 +1755,4 @@ static struct pci_driver donauboe_pci_driver = { .resume = toshoboe_wakeup }; -static int __init -donauboe_init (void) -{ - return pci_register_driver(&donauboe_pci_driver); -} - -static void __exit -donauboe_cleanup (void) -{ - pci_unregister_driver(&donauboe_pci_driver); -} - -module_init(donauboe_init); -module_exit(donauboe_cleanup); +module_pci_driver(donauboe_pci_driver); diff --git a/drivers/net/irda/vlsi_ir.c b/drivers/net/irda/vlsi_ir.c index 485006604bbc..58ef59469dd0 100644 --- a/drivers/net/irda/vlsi_ir.c +++ b/drivers/net/irda/vlsi_ir.c @@ -485,13 +485,13 @@ static int vlsi_create_hwif(vlsi_irda_dev_t *idev) idev->virtaddr = NULL; idev->busaddr = 0; - ringarea = pci_alloc_consistent(idev->pdev, HW_RING_AREA_SIZE, &idev->busaddr); + ringarea = pci_zalloc_consistent(idev->pdev, HW_RING_AREA_SIZE, + &idev->busaddr); if (!ringarea) { IRDA_ERROR("%s: insufficient memory for descriptor rings\n", __func__); goto out; } - memset(ringarea, 0, HW_RING_AREA_SIZE); hwmap = (struct ring_descr_hw *)ringarea; idev->rx_ring = vlsi_alloc_ring(idev->pdev, hwmap, ringsize[1], diff --git a/drivers/net/wireless/ipw2x00/ipw2100.c b/drivers/net/wireless/ipw2x00/ipw2100.c index dfc6dfc56d52..1ab8e500fb77 100644 --- a/drivers/net/wireless/ipw2x00/ipw2100.c +++ b/drivers/net/wireless/ipw2x00/ipw2100.c @@ -3449,8 +3449,9 @@ static int ipw2100_msg_allocate(struct ipw2100_priv *priv) return -ENOMEM; for (i = 0; i < IPW_COMMAND_POOL_SIZE; i++) { - v = pci_alloc_consistent(priv->pci_dev, - sizeof(struct ipw2100_cmd_header), &p); + v = pci_zalloc_consistent(priv->pci_dev, + sizeof(struct ipw2100_cmd_header), + &p); if (!v) { printk(KERN_ERR DRV_NAME ": " "%s: PCI alloc failed for msg " @@ -3459,8 +3460,6 @@ static int ipw2100_msg_allocate(struct ipw2100_priv *priv) break; } - memset(v, 0, sizeof(struct ipw2100_cmd_header)); - priv->msg_buffers[i].type = COMMAND; priv->msg_buffers[i].info.c_struct.cmd = (struct ipw2100_cmd_header *)v; @@ -4336,16 +4335,12 @@ static int status_queue_allocate(struct ipw2100_priv *priv, int entries) IPW_DEBUG_INFO("enter\n"); q->size = entries * sizeof(struct ipw2100_status); - q->drv = - (struct ipw2100_status *)pci_alloc_consistent(priv->pci_dev, - q->size, &q->nic); + q->drv = pci_zalloc_consistent(priv->pci_dev, q->size, &q->nic); if (!q->drv) { IPW_DEBUG_WARNING("Can not allocate status queue.\n"); return -ENOMEM; } - memset(q->drv, 0, q->size); - IPW_DEBUG_INFO("exit\n"); return 0; @@ -4374,13 +4369,12 @@ static int bd_queue_allocate(struct ipw2100_priv *priv, q->entries = entries; q->size = entries * sizeof(struct ipw2100_bd); - q->drv = pci_alloc_consistent(priv->pci_dev, q->size, &q->nic); + q->drv = pci_zalloc_consistent(priv->pci_dev, q->size, &q->nic); if (!q->drv) { IPW_DEBUG_INFO ("can't allocate shared memory for buffer descriptors\n"); return -ENOMEM; } - memset(q->drv, 0, q->size); IPW_DEBUG_INFO("exit\n"); diff --git a/drivers/net/wireless/mwl8k.c b/drivers/net/wireless/mwl8k.c index 9a3d4d6724f7..fc6cb215e761 100644 --- a/drivers/net/wireless/mwl8k.c +++ b/drivers/net/wireless/mwl8k.c @@ -1159,12 +1159,11 @@ static int mwl8k_rxq_init(struct ieee80211_hw *hw, int index) size = MWL8K_RX_DESCS * priv->rxd_ops->rxd_size; - rxq->rxd = pci_alloc_consistent(priv->pdev, size, &rxq->rxd_dma); + rxq->rxd = pci_zalloc_consistent(priv->pdev, size, &rxq->rxd_dma); if (rxq->rxd == NULL) { wiphy_err(hw->wiphy, "failed to alloc RX descriptors\n"); return -ENOMEM; } - memset(rxq->rxd, 0, size); rxq->buf = kcalloc(MWL8K_RX_DESCS, sizeof(*rxq->buf), GFP_KERNEL); if (rxq->buf == NULL) { @@ -1451,12 +1450,11 @@ static int mwl8k_txq_init(struct ieee80211_hw *hw, int index) size = MWL8K_TX_DESCS * sizeof(struct mwl8k_tx_desc); - txq->txd = pci_alloc_consistent(priv->pdev, size, &txq->txd_dma); + txq->txd = pci_zalloc_consistent(priv->pdev, size, &txq->txd_dma); if (txq->txd == NULL) { wiphy_err(hw->wiphy, "failed to alloc TX descriptors\n"); return -ENOMEM; } - memset(txq->txd, 0, size); txq->skb = kcalloc(MWL8K_TX_DESCS, sizeof(*txq->skb), GFP_KERNEL); if (txq->skb == NULL) { diff --git a/drivers/net/wireless/rtl818x/rtl8180/dev.c b/drivers/net/wireless/rtl818x/rtl8180/dev.c index 4b904f708184..fcc45e5bf50a 100644 --- a/drivers/net/wireless/rtl818x/rtl8180/dev.c +++ b/drivers/net/wireless/rtl818x/rtl8180/dev.c @@ -972,16 +972,13 @@ static int rtl8180_init_rx_ring(struct ieee80211_hw *dev) else priv->rx_ring_sz = sizeof(struct rtl8180_rx_desc); - priv->rx_ring = pci_alloc_consistent(priv->pdev, - priv->rx_ring_sz * 32, - &priv->rx_ring_dma); - + priv->rx_ring = pci_zalloc_consistent(priv->pdev, priv->rx_ring_sz * 32, + &priv->rx_ring_dma); if (!priv->rx_ring || (unsigned long)priv->rx_ring & 0xFF) { wiphy_err(dev->wiphy, "Cannot allocate RX ring\n"); return -ENOMEM; } - memset(priv->rx_ring, 0, priv->rx_ring_sz * 32); priv->rx_idx = 0; for (i = 0; i < 32; i++) { @@ -1040,14 +1037,14 @@ static int rtl8180_init_tx_ring(struct ieee80211_hw *dev, dma_addr_t dma; int i; - ring = pci_alloc_consistent(priv->pdev, sizeof(*ring) * entries, &dma); + ring = pci_zalloc_consistent(priv->pdev, sizeof(*ring) * entries, + &dma); if (!ring || (unsigned long)ring & 0xFF) { wiphy_err(dev->wiphy, "Cannot allocate TX ring (prio = %d)\n", prio); return -ENOMEM; } - memset(ring, 0, sizeof(*ring)*entries); priv->tx_ring[prio].desc = ring; priv->tx_ring[prio].dma = dma; priv->tx_ring[prio].idx = 0; diff --git a/drivers/net/wireless/rtlwifi/pci.c b/drivers/net/wireless/rtlwifi/pci.c index dae55257f0e8..67d1ee6edcad 100644 --- a/drivers/net/wireless/rtlwifi/pci.c +++ b/drivers/net/wireless/rtlwifi/pci.c @@ -1092,16 +1092,14 @@ static int _rtl_pci_init_tx_ring(struct ieee80211_hw *hw, u32 nextdescaddress; int i; - ring = pci_alloc_consistent(rtlpci->pdev, - sizeof(*ring) * entries, &dma); - + ring = pci_zalloc_consistent(rtlpci->pdev, sizeof(*ring) * entries, + &dma); if (!ring || (unsigned long)ring & 0xFF) { RT_TRACE(rtlpriv, COMP_ERR, DBG_EMERG, "Cannot allocate TX ring (prio = %d)\n", prio); return -ENOMEM; } - memset(ring, 0, sizeof(*ring) * entries); rtlpci->tx_ring[prio].desc = ring; rtlpci->tx_ring[prio].dma = dma; rtlpci->tx_ring[prio].idx = 0; @@ -1139,10 +1137,9 @@ static int _rtl_pci_init_rx_ring(struct ieee80211_hw *hw) for (rx_queue_idx = 0; rx_queue_idx < RTL_PCI_MAX_RX_QUEUE; rx_queue_idx++) { rtlpci->rx_ring[rx_queue_idx].desc = - pci_alloc_consistent(rtlpci->pdev, - sizeof(*rtlpci->rx_ring[rx_queue_idx]. - desc) * rtlpci->rxringcount, - &rtlpci->rx_ring[rx_queue_idx].dma); + pci_zalloc_consistent(rtlpci->pdev, + sizeof(*rtlpci->rx_ring[rx_queue_idx].desc) * rtlpci->rxringcount, + &rtlpci->rx_ring[rx_queue_idx].dma); if (!rtlpci->rx_ring[rx_queue_idx].desc || (unsigned long)rtlpci->rx_ring[rx_queue_idx].desc & 0xFF) { @@ -1151,10 +1148,6 @@ static int _rtl_pci_init_rx_ring(struct ieee80211_hw *hw) return -ENOMEM; } - memset(rtlpci->rx_ring[rx_queue_idx].desc, 0, - sizeof(*rtlpci->rx_ring[rx_queue_idx].desc) * - rtlpci->rxringcount); - rtlpci->rx_ring[rx_queue_idx].idx = 0; /* If amsdu_8k is disabled, set buffersize to 4096. This diff --git a/drivers/net/xen-netback/common.h b/drivers/net/xen-netback/common.h index ef3026f46a37..d4eb8d2e9cb7 100644 --- a/drivers/net/xen-netback/common.h +++ b/drivers/net/xen-netback/common.h @@ -165,6 +165,7 @@ struct xenvif_queue { /* Per-queue data for xenvif */ u16 dealloc_ring[MAX_PENDING_REQS]; struct task_struct *dealloc_task; wait_queue_head_t dealloc_wq; + atomic_t inflight_packets; /* Use kthread for guest RX */ struct task_struct *task; @@ -329,4 +330,8 @@ extern unsigned int xenvif_max_queues; extern struct dentry *xen_netback_dbg_root; #endif +void xenvif_skb_zerocopy_prepare(struct xenvif_queue *queue, + struct sk_buff *skb); +void xenvif_skb_zerocopy_complete(struct xenvif_queue *queue); + #endif /* __XEN_NETBACK__COMMON_H__ */ diff --git a/drivers/net/xen-netback/interface.c b/drivers/net/xen-netback/interface.c index bfd10cb9c8de..e29e15dca86e 100644 --- a/drivers/net/xen-netback/interface.c +++ b/drivers/net/xen-netback/interface.c @@ -43,6 +43,23 @@ #define XENVIF_QUEUE_LENGTH 32 #define XENVIF_NAPI_WEIGHT 64 +/* This function is used to set SKBTX_DEV_ZEROCOPY as well as + * increasing the inflight counter. We need to increase the inflight + * counter because core driver calls into xenvif_zerocopy_callback + * which calls xenvif_skb_zerocopy_complete. + */ +void xenvif_skb_zerocopy_prepare(struct xenvif_queue *queue, + struct sk_buff *skb) +{ + skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY; + atomic_inc(&queue->inflight_packets); +} + +void xenvif_skb_zerocopy_complete(struct xenvif_queue *queue) +{ + atomic_dec(&queue->inflight_packets); +} + static inline void xenvif_stop_queue(struct xenvif_queue *queue) { struct net_device *dev = queue->vif->dev; @@ -524,9 +541,6 @@ int xenvif_init_queue(struct xenvif_queue *queue) init_timer(&queue->rx_stalled); - netif_napi_add(queue->vif->dev, &queue->napi, xenvif_poll, - XENVIF_NAPI_WEIGHT); - return 0; } @@ -560,6 +574,7 @@ int xenvif_connect(struct xenvif_queue *queue, unsigned long tx_ring_ref, init_waitqueue_head(&queue->wq); init_waitqueue_head(&queue->dealloc_wq); + atomic_set(&queue->inflight_packets, 0); if (tx_evtchn == rx_evtchn) { /* feature-split-event-channels == 0 */ @@ -614,6 +629,9 @@ int xenvif_connect(struct xenvif_queue *queue, unsigned long tx_ring_ref, wake_up_process(queue->task); wake_up_process(queue->dealloc_task); + netif_napi_add(queue->vif->dev, &queue->napi, xenvif_poll, + XENVIF_NAPI_WEIGHT); + return 0; err_rx_unbind: @@ -642,25 +660,6 @@ void xenvif_carrier_off(struct xenvif *vif) rtnl_unlock(); } -static void xenvif_wait_unmap_timeout(struct xenvif_queue *queue, - unsigned int worst_case_skb_lifetime) -{ - int i, unmap_timeout = 0; - - for (i = 0; i < MAX_PENDING_REQS; ++i) { - if (queue->grant_tx_handle[i] != NETBACK_INVALID_HANDLE) { - unmap_timeout++; - schedule_timeout(msecs_to_jiffies(1000)); - if (unmap_timeout > worst_case_skb_lifetime && - net_ratelimit()) - netdev_err(queue->vif->dev, - "Page still granted! Index: %x\n", - i); - i = -1; - } - } -} - void xenvif_disconnect(struct xenvif *vif) { struct xenvif_queue *queue = NULL; @@ -672,6 +671,8 @@ void xenvif_disconnect(struct xenvif *vif) for (queue_index = 0; queue_index < num_queues; ++queue_index) { queue = &vif->queues[queue_index]; + netif_napi_del(&queue->napi); + if (queue->task) { del_timer_sync(&queue->rx_stalled); kthread_stop(queue->task); @@ -704,7 +705,6 @@ void xenvif_disconnect(struct xenvif *vif) void xenvif_deinit_queue(struct xenvif_queue *queue) { free_xenballooned_pages(MAX_PENDING_REQS, queue->mmap_pages); - netif_napi_del(&queue->napi); } void xenvif_free(struct xenvif *vif) @@ -712,21 +712,11 @@ void xenvif_free(struct xenvif *vif) struct xenvif_queue *queue = NULL; unsigned int num_queues = vif->num_queues; unsigned int queue_index; - /* Here we want to avoid timeout messages if an skb can be legitimately - * stuck somewhere else. Realistically this could be an another vif's - * internal or QDisc queue. That another vif also has this - * rx_drain_timeout_msecs timeout, so give it time to drain out. - * Although if that other guest wakes up just before its timeout happens - * and takes only one skb from QDisc, it can hold onto other skbs for a - * longer period. - */ - unsigned int worst_case_skb_lifetime = (rx_drain_timeout_msecs/1000); unregister_netdev(vif->dev); for (queue_index = 0; queue_index < num_queues; ++queue_index) { queue = &vif->queues[queue_index]; - xenvif_wait_unmap_timeout(queue, worst_case_skb_lifetime); xenvif_deinit_queue(queue); } diff --git a/drivers/net/xen-netback/netback.c b/drivers/net/xen-netback/netback.c index 4734472aa620..08f65996534c 100644 --- a/drivers/net/xen-netback/netback.c +++ b/drivers/net/xen-netback/netback.c @@ -1525,10 +1525,12 @@ static int xenvif_handle_frag_list(struct xenvif_queue *queue, struct sk_buff *s /* remove traces of mapped pages and frag_list */ skb_frag_list_init(skb); uarg = skb_shinfo(skb)->destructor_arg; + /* increase inflight counter to offset decrement in callback */ + atomic_inc(&queue->inflight_packets); uarg->callback(uarg, true); skb_shinfo(skb)->destructor_arg = NULL; - skb_shinfo(nskb)->tx_flags |= SKBTX_DEV_ZEROCOPY; + xenvif_skb_zerocopy_prepare(queue, nskb); kfree_skb(nskb); return 0; @@ -1589,7 +1591,7 @@ static int xenvif_tx_submit(struct xenvif_queue *queue) if (net_ratelimit()) netdev_err(queue->vif->dev, "Not enough memory to consolidate frag_list!\n"); - skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY; + xenvif_skb_zerocopy_prepare(queue, skb); kfree_skb(skb); continue; } @@ -1609,7 +1611,7 @@ static int xenvif_tx_submit(struct xenvif_queue *queue) "Can't setup checksum in net_tx_action\n"); /* We have to set this flag to trigger the callback */ if (skb_shinfo(skb)->destructor_arg) - skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY; + xenvif_skb_zerocopy_prepare(queue, skb); kfree_skb(skb); continue; } @@ -1641,7 +1643,7 @@ static int xenvif_tx_submit(struct xenvif_queue *queue) * skb. E.g. the __pskb_pull_tail earlier can do such thing. */ if (skb_shinfo(skb)->destructor_arg) { - skb_shinfo(skb)->tx_flags |= SKBTX_DEV_ZEROCOPY; + xenvif_skb_zerocopy_prepare(queue, skb); queue->stats.tx_zerocopy_sent++; } @@ -1681,6 +1683,7 @@ void xenvif_zerocopy_callback(struct ubuf_info *ubuf, bool zerocopy_success) queue->stats.tx_zerocopy_success++; else queue->stats.tx_zerocopy_fail++; + xenvif_skb_zerocopy_complete(queue); } static inline void xenvif_tx_dealloc_action(struct xenvif_queue *queue) @@ -2058,15 +2061,24 @@ int xenvif_kthread_guest_rx(void *data) return 0; } +static bool xenvif_dealloc_kthread_should_stop(struct xenvif_queue *queue) +{ + /* Dealloc thread must remain running until all inflight + * packets complete. + */ + return kthread_should_stop() && + !atomic_read(&queue->inflight_packets); +} + int xenvif_dealloc_kthread(void *data) { struct xenvif_queue *queue = data; - while (!kthread_should_stop()) { + for (;;) { wait_event_interruptible(queue->dealloc_wq, tx_dealloc_work_todo(queue) || - kthread_should_stop()); - if (kthread_should_stop()) + xenvif_dealloc_kthread_should_stop(queue)); + if (xenvif_dealloc_kthread_should_stop(queue)) break; xenvif_tx_dealloc_action(queue); diff --git a/drivers/net/xen-netback/xenbus.c b/drivers/net/xen-netback/xenbus.c index 580517d857bf..9c47b897b6d2 100644 --- a/drivers/net/xen-netback/xenbus.c +++ b/drivers/net/xen-netback/xenbus.c @@ -116,6 +116,7 @@ static int xenvif_read_io_ring(struct seq_file *m, void *v) } #define XENVIF_KICK_STR "kick" +#define BUFFER_SIZE 32 static ssize_t xenvif_write_io_ring(struct file *filp, const char __user *buf, size_t count, @@ -124,22 +125,24 @@ xenvif_write_io_ring(struct file *filp, const char __user *buf, size_t count, struct xenvif_queue *queue = ((struct seq_file *)filp->private_data)->private; int len; - char write[sizeof(XENVIF_KICK_STR)]; + char write[BUFFER_SIZE]; /* don't allow partial writes and check the length */ if (*ppos != 0) return 0; - if (count < sizeof(XENVIF_KICK_STR) - 1) + if (count >= sizeof(write)) return -ENOSPC; len = simple_write_to_buffer(write, - sizeof(write), + sizeof(write) - 1, ppos, buf, count); if (len < 0) return len; + write[len] = '\0'; + if (!strncmp(write, XENVIF_KICK_STR, sizeof(XENVIF_KICK_STR) - 1)) xenvif_interrupt(0, (void *)queue); else { @@ -171,10 +174,9 @@ static const struct file_operations xenvif_dbg_io_ring_ops_fops = { .write = xenvif_write_io_ring, }; -static void xenvif_debugfs_addif(struct xenvif_queue *queue) +static void xenvif_debugfs_addif(struct xenvif *vif) { struct dentry *pfile; - struct xenvif *vif = queue->vif; int i; if (IS_ERR_OR_NULL(xen_netback_dbg_root)) @@ -733,10 +735,11 @@ static void connect(struct backend_info *be) be->vif->num_queues = queue_index; goto err; } + } + #ifdef CONFIG_DEBUG_FS - xenvif_debugfs_addif(queue); + xenvif_debugfs_addif(be->vif); #endif /* CONFIG_DEBUG_FS */ - } /* Initialisation completed, tell core driver the number of * active queues. |